> ## Documentation Index
> Fetch the complete documentation index at: https://larkuprag.larkup.de/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Deploy Your Server
> Ship your RAG pipeline to production — choose your preferred cloud provider.
LarkupRAG generates a fully self-contained Node.js server with zero bloat. Because it's standard Node.js under the hood, **you can deploy it anywhere** — serverless platforms, bare-metal VPS, or managed cloud services.
Every generated server ships with a `vercel.json` for Vercel and a `Dockerfile` for VPS / container deployments out of the box. No extra configuration needed.
***
## Choose Your Deployment Target
Recommended · Serverless
The fastest path to production. The generated server includes a `vercel.json` pre-configured for serverless deployment — one command and you're live.
### Demo
> **Note:** Full demo video coming soon. Screenshots below show the in-app deployment flow.
### Steps
From the **Server** page in the Larkup-RAG app, click the **Deploy** button in the top-right corner and select **Vercel** from the dropdown.
Paste your Vercel Access Token into the field. You can generate one at [vercel.com/account/tokens](https://vercel.com/account/tokens).
Enter a unique **Project ID or Name** (e.g. `my-rag-server`). This is saved per server for future re-deployments.
Click **Configure** next to the Environment Variables section. Your workspace `.env` values are pre-filled automatically. Ensure at minimum your LLM provider API key is set (e.g. `OPENAI_API_KEY`).
Click **Deploy** — Larkup-RAG will push your server to Vercel. Once complete, a live URL will appear in the Server panel.
After your first deployment, subsequent deploys are instant — just click Deploy again. Your project name and token are remembered.
Docker · SSH Deploy
Deploy to any Linux VPS (Hetzner, DigitalOcean Droplets, AWS EC2, etc.) via SSH. Larkup-RAG automatically connects, installs Docker, and launches your server — no manual setup needed.
### Demo
> **Note:** Full demo video coming soon. Screenshots below show the live SSH deployment in action.
### Steps
Create a fresh Ubuntu 22.04+ server on [Hetzner Cloud](https://www.hetzner.com/cloud), DigitalOcean, or any VPS provider. Note the public IP address.
In the Larkup-RAG app, click **Deploy → Deploy via SSH (Hetzner, EC2, Droplets)**.
Fill in:
* **Server IP / Hostname** — e.g. `167.233.113.67`
* **SSH Username** — typically `root` for a fresh VPS
* **Authentication Type** — choose **Password** or **Private Key**
Click **Configure** to set your API keys. Values from your local `.env` are pre-filled and synced automatically.
Click **Deploy via SSH**. Larkup-RAG will:
1. Connect to your server over SSH
2. Install Docker (if not present)
3. Pull and build your RAG server image
4. Launch the container on port `8080`
You can watch the live terminal output stream directly in the app.
Once deployment completes, the live server URL (`http://:8080`) appears in the panel with a ready-to-use `curl` test command.
For production use, configure a reverse proxy (Nginx or Caddy) in front of port `8080` to handle SSL termination and custom domain routing.
Azure App Service · Container Apps
LarkupRAG works seamlessly on Microsoft Azure using either **Azure App Service** (for Node.js apps) or **Azure Container Apps** (for Dockerized workloads).
A detailed step-by-step Azure walkthrough with screenshots is coming soon. In the meantime, reach out via [GitHub Issues](https://github.com/Larkup-AI/larkup-rag/issues) if you need help.
AWS · GCP · Railway · Fly.io · and more
LarkupRAG's generated server is a plain Node.js app with an included `Dockerfile` — it runs on **any cloud or platform** that supports containers or Node.js.
### Supported Platforms
| Platform | Method | Notes |
| --------------------- | ------------ | -------------------------------------------------------- |
| **AWS EC2** | SSH / Docker | Use the Hetzner SSH deploy flow with your EC2 IP |
| **AWS ECS / Fargate** | Docker | Push to ECR, deploy as ECS task |
| **AWS App Runner** | Docker / Git | Connect your repo or push an image directly |
| **Google Cloud Run** | Docker | Push to Artifact Registry, deploy as a Cloud Run service |
| **Railway** | Git / Docker | `railway up` from the generated server directory |
| **Fly.io** | Docker | `fly launch` from the generated server directory |
| **Render** | Docker / Git | Connect your repo or use the `Dockerfile` |
| **Self-hosted** | Docker | Run anywhere with `docker run` |
### Generic Docker Deploy
All platforms that support containers can use the included `Dockerfile`:
```bash theme={null}
# Build
docker build -t my-rag-server .
# Run (replace env vars as needed)
docker run -d \
-p 8080:8080 \
-e OPENAI_API_KEY=your_key \
-e SERVER_API_KEY=your_secret \
my-rag-server
```
### AWS-Specific Quick Start
```bash theme={null}
aws ecr create-repository --repository-name my-rag-server
aws ecr get-login-password | docker login --username AWS \
--password-stdin .dkr.ecr..amazonaws.com
docker tag my-rag-server:latest \
.dkr.ecr..amazonaws.com/my-rag-server:latest
docker push \
.dkr.ecr..amazonaws.com/my-rag-server:latest
```
Create a task definition or App Runner service pointing to your ECR image. Set environment variables in the service configuration.
Need first-class support for a specific cloud? [Open an issue](https://github.com/Larkup-AI/larkup-rag/issues) or submit a PR — one-click deploy integrations for AWS, GCP, and Railway are on the roadmap.
***
## Environment Variables Reference
Regardless of your deployment target, ensure the following variables are configured in your cloud provider's settings panel:
| Variable | Required | Description |
| ------------------- | ----------------------- | ---------------------------------------------------- |
| `OPENAI_API_KEY` | ✅ Yes (if using OpenAI) | Your OpenAI API key for embeddings / completions |
| `SERVER_API_KEY` | Recommended | Secures your deployed RAG endpoints with Bearer auth |
| `EMBEDDING_API_KEY` | ✅ Yes | API key used to embed incoming queries |
| `LANCEDB_TABLE` | Optional | Vector table name (default: `documents`) |
| `LANCEDB_URI` | Optional | LanceDB Cloud URI (for cloud vector storage mode) |
| `LANCEDB_API_KEY` | Optional | LanceDB Cloud API key |